Astronomy&Astrophysicsmanuscriptno.stasinska-3533 February5,2008 (DOI:willbeinsertedbyhandlater) Post-AGB stars as testbeds of nucleosynthesis in AGB stars G.Stasin´ska1,R.Szczerba2,M.Schmidt2,andN.Sio´dmiak2 1 LUTH,ObservatoiredeMeudon,5PlaceJulesJanssen,F-92195MeudonCedex,France 2 N.CopernicusAstronomicalCenter,Rabian´ska8,87-100Torun´,Poland 6 Received/Accepted 0 0 2 Abstract. We construct a data base of 125 post-AGB objects (including R CrB and extreme helium stars) with published photospheric parameters(effectivetemperature andgravity) andchemical composition. Weestimatethemassesof thepost- n AGBstarsbycomparingtheirpositioninthe(logT ,logg)planewiththeoreticalevolutionarytracksofdifferentmasses.We a eff J constructvariousdiagrams,withtheaimoffindingcluestoAGBnucleosynthesis.Thisisthefirsttimethatalargesampleof post-AGBstarshasbeenusedinasystematicwayforsuchapurposeandwearguethat,inseveralrespects,post-AGBstars 3 2 should be morepowerful than planetary nebulaetotest AGB nucleosynthesis. Our mainfindings arethat: thevast majority ofobjectswhichdonotshowevidenceofNproductionfromprimaryChavealowstellarmass(M⋆ <0.56M⊙);thereisno 1 evidencethatobjectswhichdidnotexperience3rddredge-up haveadifferentstellarmassdistributionthanobjectsthatdid; v thereisclearevidencethat3rddredge-upismoreefficientatlowmetallicity.Thesampleofknownpost-AGBstarsislikelyto 4 increasesignificantlyinthenearfuturethankstotheASTRO-Fandfollow-upobservations,makingtheseobjectsevenmore 0 promisingastestbedsforAGBnucleosynthesis. 5 1 Keywords.stars:AGBandpostAGBstars—Stars:abundances—Stars:evolution—Physicaldataandprocesses:Nuclear 0 reactions,nucleosynthesis,abundances 6 0 / h 1. Introduction to the planetary nebula ejection. With the availability of fast p computers,itisnowpossibletocomputecompleteAGBmod- - o The asymptotic giantbranch (AGB) phase, a late stage in the els(Karakasetal.2002,Herwig2004)thatfollowallthepulses r evolution of low and intermediate mass stars, is quite com- t indetail.Testingmodelpredictionsbeforeusingtheminchem- s plex.Duringthisphase,variousnuclearprocessesareatwork ical evolution models of galaxies is vital, especially because a in different zones of the star, and a variety of mixing mecha- : even full evolutionary calculations are computationally diffi- v nisms take place (see e.g. Charbonnel2002 or Lattanzio2002 cult and still imply some ad hoc parameters (mixing length, i forshortreviews).Understandingthisphaseisimportant,since X mass loss ratesetc.). Indeed,as shown recentlybyVentura& theelementsmanufacturedduringtheAGBcontributesignifi- r D’Antona (2005a, 2005b), the predicted yields of intermedi- a cantlytothechemicalcompositionofgalaxies. ate mass stars depend strongly on the treatment adopted for The first models to give some predictions on the stellar convection and mass loss and on the nuclear reaction cross- yields from AGB stars are those by Iben & Truran(1978) sections. andRenzini&Voli(1981).Unfortunately,evennowadays,the As mentioned before, synthetic models published since state of stellar physics does not allow one to construct mod- 1993 reproduce some observables (e.g. the luminosity func- els entirely from first principles, and some quantities have to tionsofcarbon-starsandoflithium-richstars)byconstruction. be set as free parameters. The next generation of AGB mod- However, additional tests are needed and are crucial to con- els (Groenewegen & de Jong1993, Marigo et al.1996) ad- strainAGBmodels.Theanalysisofthechemicalcomposition justed those free parameters (essentially mass loss rate and ofplanetarynebulaecanprovidesometests.Suchanapproach mixing length) to reproduce a few observational constraints hasbeenadoptedrecentlybyMarigoetal.(2003)usingadata on stellar populations. Further models have been constructed baseof10planetarynebulae(PNe)observedinawidespectral since then (e.g. by Forestini & Charbonnel1997, Boothroyd range, from the far infrared to the ultraviolet. Previously, the &Sackmann1999,Marigo2001,Izzardetal.2004).Allthose chemical composition of planetary nebulae samples had been modelsare so-called synthetic modelsin that, for the thermal usedinamoreempiricalwaytotestAGBnucleosynthesisand pulsephase,theyuseanalyticalexpressionstoextrapolatecer- getaninsightintotherelationbetweenplanetarynebulae,the tainquantitiesobtainedfromfullevolutionarycalculationsup finalproductsoflowandintermediatemassstarsandtheirpro- Sendoffprintrequeststo:G.Stasin´ska, genitors. For example, Peimbert(1978) defined a category of e-mail:[email protected] planetarynebulae,called TypeI PNe, as objectshavingHe/H 2 Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars > 0.125andN/O > 0.5.Those PNe were interpretedasbeing 2. Theopticalsampleofpost-AGBstars. bornfromstarsthatexperiencedtheseconddredge-up.While the designation”TypeI PNe” stayed in the literature,its defi- 2.1.Generalpresentation nitionchangedseveraltimes(seeStasin´ska2004).Kingsburgh Asmentionedabove,thepost-AGBphasestartswhenthestar & Barlow(1994) comparedthe nebular N/H ratio to the solar hasexpelleditsenvelopeandlefttheAGBbranch,andisthen (C+N)/H ratio to find out which planetary nebulae exhibit N movingtotheleftintheH-Rdiagram.TheendofAGBischar- producedfromprimaryC.Theyalso identifiedplanetaryneb- acterized by strong mass loss, which (sometimes) can reach ulaewhoseprogenitorshadexperienced3rddredge-up,bring- ingtothesurfacecarbonproducedbythetriple-αreaction.To valuesof10−4 M⊙yr−1.Forstarsthathaveexperiencedintense masslossontheAGB,duetothelargedustopacityintheenve- thisend,theyconstructeda(C+N+O)/HvsC/Hdiagram.The lope,thepost-AGBstarisfirstseenthroughitsinfraredemis- N/O vs O/H diagram has been interpreted (e.g. Henry1990) sionduetoreprocessingofthestellarradiationbythecircum- as indicating that in some objects N is produced at the ex- stellardustgrains.Itisonlywhentheenvelopehassufficiently pense of O (during ON cycling). However, this diagram, de- expandedandbecomeopticallythinthatsuchstarsstartbeing pendingonauthorsandsamples,doesnotalwaysleadtosuch optically visible. In this paper, we are interested in optically a straightforward interpretation (Kingsburgh & Barlow1994, visiblepost-AGBstars.Weusedthepresentversion(Szczerba Leisy&Dennefeld1996).Henryetal.(2000)haveplottedPN etal.,inpreparation)ofourcatalogueofpost-AGBcandidates abundancesof C and N as a function of progenitormass (es- (Szczerbaetal.2001),whichnowcontainsabout330objects. timated from the stellar remnantmasses by Go´rny et al.1997 For all the sources from this catalogue we have searched the and Stasin´ska et al.1997 and converted to progenitor masses availableliteratureforstellarparametersandchemicalcompo- usinganinitial-finalmassrelation),andcomparedthiswiththe sition. predictionsfromsyntheticAGBmodels.Thiswasthefirstat- tempttocomparePNe abundancesdirectlywiththeresultsof We then estimated the stellar masses, M⋆, by comparison AGB model computations for a given initial mass and it was withtheoreticalevolutionarypathsinthe(logTeff,logg)plane. notveryconclusive. The theoretical paths we used were interpolated by Go´rny et al.(1997) from the post-AGB models of Scho¨nberner(1983) Thereisanothercategoryofobjectsthatmayservefortests and Blo¨cker(1995). The uncertaintiesin logTeff and logg in- ofAGBmodels.Thesearepost-AGBstars,i.e.starsthathave duce an uncertainty in the derived M⋆. The main source of already ejected their envelope but are not yet hot enough to uncertaintiesisthelowaccuracyofthedeterminationsofsur- ionizeitandproducea planetarynebula.Suchstarshavesev- facegravity,especiallyinthecaseofcoolpost-AGBstars(Teff eraladvantageswithrespecttoPNeandconstituteanexcellent below 10,000K). Since the relation between logg and stel- complementfortestingAGBnucleosynthesis.Themainadvan- lar mass on the post-AGB tracks is highly nonlinear,for each tage is that the stellar mass can be obtained directly from the object we determined the mass M⋆min corresponding to (log observed stellar spectrum by fitting a model atmosphere that Teff − ∆(logTeff), log g − ∆(logg)) and the mass M⋆max cor- allowsonetoderivethestellareffectivetemperature,Teff,and responding to (log Teff +∆(logTeff), log g +∆(llogg)). Since gravityg(thiscanalsobedoneforPNenuclei,butasthestars the derived stellar mass is a decreasing function of both Teff hotterandburiedintheionizedgas,thisismoredifficult).The andlogg,thevaluesof M⋆min and M⋆max shoulddefineacon- abundancesofquiteavarietyofelements(He,Li,C,N,O,Na, servativestellar mass interval.1 However,oneshould keepin Mg,Al,Si,S, Ca,Sc,Ti, Fe,Ni,Zn...)canbeobtainedfrom mind that many post-AGBstars are pulsating stars (Gautschy astellaratmosphereanalysis.Forcarbon,aparticularlyimpor- & Saio1996), while such pulses are not reproduced by the tantelementinAGBevolution,theuncertaintyintheestimated evolutionary models used to derive the stellar masses. In ad- abundanceisofthesameorderasfortheotherelements.Thisis dition, the pulsating nature of the atmosphere may introduce notthecaseinplanetarynebulae,wherenostrongcarbonline errorsin the abundancedeterminations,which are doneusing existsintheoptical,implyingthatthecarbonabundancedeter- static atmosphere models. A further source of uncertainty in minationissubjecttoa higheruncertaintythantheoxygenor thedeterminationofmassescomes,ofcourse,fromthemodel thenitrogenabundancedetermination.Finally,post-AGBstars tracksthemselves.Allthisimpliesthatthemassesofpost-AGB are expected to extend to smaller masses than the nuclei of starsarequiteuncertainandmodel-dependent.Inspiteofthis, planetary nebulae. Indeed,planetary nebulae with central star webelievethatthemethodprovidesusefulinformation.Itcan masses lowerthan0.55M⊙ do notexistsince the nebulargas be improvedon in the future, as better spectra become avail- has dissipated in the interstellar medium long before the star ableforspectroscopicanalysisandasprogressinunderstand- hasbecomehotenoughtoionizeit.Ontheotherhand,thereis ingthephysicsofpost-AGBstarsismade.Inparticular,when nosuchlimitationforpost-AGBstars. completegridswithdifferentmetallicitiesbecomeavailable,it shouldbepossibletoaccountforthemetallicityinthederiva- This paperpresentsthe first attemptto use a largesample tion of the stellar masses. For the moment, as far as one can ofpost-AGBstarstotestAGBnucleosynthesis.InSect.2,we judge from several tracks computed by Vassiliadis & Wood describe the constitution of our sample. In Sect. 3 we show (1994)atdifferentmetallicities,theeffectofmetallicityinmass some empirical diagrams and propose simple interpretations. In Sect. 4 we summarize the main findings and outline some 1 The available model grid allows us to determine masses only prospects. withintherange(0.55–0.94M⊙). Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars 3 derivationwouldbecompletelydominatedbythelargeerroron sources for the abundances are the same as given in col. (8) logg. ofTable1.Overall,thetypicaluncertaintyinelementalabun- The details of our search and the above estimates of the dancesisabout0.2-0.3dex.However,theuncertaintyinabun- stellarmassesarepresentedinTable1.Theobjectsaregrouped danceratiosofheavyelementsissmaller,sincemanysources intoseveralsubtypes:RVTaustars(29objects),suspectedRV of errors affect the derived abundancesin a similar way. One Taustars(6objectsfromMaasetal.2005),RCrBstars(19ob- exceptionisthecase ofoxygen,if itsabundancehasbeende- jects),extremeheliumstars(15objects)andalltheremaining terminedfromtheO7771-5triplet.Itiswellknownthatthis post-AGBstars(56objects). triplet gives enhanced abundances,if non-LTE effects are not TheRVTaustarsarehighlyluminousvariablestarscharac- taken into account, and that the O 7771-5vs [O ] discrep- terizedbylight-curveswithalternatedeepandshallowminima, ancy is higher for low metallicity (Takeda2003). In most ob- periodsbetween30and150days,andF,GorKspectraltypes jectslistedinTable2,itappearsthattheuseofthistripletwas (seee.g.Prestonetal.1963).Theyhavebeenidentifiedaspost- avoided. AGBstarsbyJura(1986),whoshowedthattheirIRASfluxes indicatethattheyhavejustleftaphaseofveryrapidmassloss. 2.2.Notesonindividualobjectsanddiscussionof TheRCrBstars(alreadyknownformorethan200years!) abundanceuncertainties arerareH-deficientandC-richsupergiantsthatundergoirreg- ulardeclinesofupto8magnitudeswhendustformsinclumps Forseveralsourceswefoundmorethanonereferencewithstel- alongthelineofsight(seee.gClayton1996forareview).The lar parameters and chemical composition determined. Below extreme H-deficiency of the R CrB stars suggests that some we presentargumentsforthepreferredsourceofinformation. mechanism removedthe entire H-rich stellar envelope.There Theobjectnumbercorrespondstothenumbergivenincolumn aretwomajormodelswhichexplaintheirorigin:amergersce- (1)ofTables1and2. nario(Webbink1984,Ibenetal.1986)orthefinalheliumshell Object 1: IRAS18384−2800. The atmospheric parameters flashscenario(Fujimoto1977,Renzini1979).Thereisstillno and chemical abundances have been analyzed recently by consensusaboutwhichscenarioisvalid(noneofthemcanex- AFGM01andRvW01.Thesameatmosphericparameterswere plainalltheobservedproperties).Onlythesecondscenarioim- derivedinbothanalyses.ThespectraobtainedbyRvW01have pliesapost-AGBnature.WehaveincludedthesestarsinTable apparentlymuchhigherS/N-ratiothanthoseofAFGM01,and 1, but we do not consider them as bona-fide post-AGB stars. thechemicalanalysisofRvW01isbasedonahighernumberof Extremeheliumstars,whichcouldbeevolutionarilyconnected lines,sotheresultsofRvW01werechosenforfurtheranalysis. toRCrBstars(see e.g.Pandeyetal.2001),arealso included Object3:IRAS17279−1119.Thisstarhasbeenanalyzedby inourTableanddiscussedtogetherwithRCrBstars. AFGM01andVW97.BothanalysesgavesimilarvalueforT Ineachsubtype,theobjectsinTable1areorderedbygalac- eff butratherdifferentvaluesforthesurfacegravity.Sincethede- ticcoordinateslandb.Forsomesources,thereareseveralen- terminationofAFGM01wasbasedonasinglespectrum,while triesandinSect.2.2webrieflydiscussthepreferreddetermina- thevW97determinationwasbasedonseveralspectraatdiffer- tions- usually,theyarebasedonhigherqualityspectroscopic entphotometricphases,itislikely thatmoreconsistentatmo- material. The columnscontain the followingdata: (1) the ob- spheric parameters and chemical composition are obtained in jectnumber;(2)theobjectcoordinatesl,b;(3)theIRASname; theAFGM01paper. (4)theHD number;(5)othername(eithertheusualnameor Object 5: IRAS19500−1709. Both analyses by vWR00 and the designation in one of the following catalogues (chosen in vWWW96 give similar atmospheric parameters. The vWR00 thisorder:GeneralCatalogueofVariableStars,LS,BD,SAO, paperisbasedonahigherS/Nandabroaderspectralcoverage andCD catalogues); (6)the effectivetempearture;(7)the er- andwaschosenforthesubsequentanalysis. ror attributed to T ; (8) the logarithm of the surface gravity incms−2;(9)theeerffrorattributedtologg;(10)referencesand Object6:IRAS19590−1249.WeadoptedtheRDM03values. TheanalysisofRDM03isbasedonhigherqualitymaterialand notesforthecollectedstellarparameters(explanationsforthe is based on fullyblanketednon-LTEatmosphericmodelsthat abbreviations used and for notes are given at the end of the shouldguaranteemoreaccuratevaluesforT andlogg. table); (11) M⋆ in units of solar mass; (12) M⋆min and (13) eff Object8:PHL1580.KL86donotgiveabundances,sowehad M⋆max. torelyonthosegivenbyCDK91. Table2isorderedinthesamewayasTable1.Itscolumns contain:(1)theobjectnumber;(2)theobjectcoordinatesl,b; Object10:IRAS19114+0002.TheanalysisofRH99isbased (3)– (8)theabundancesofC, N,O,S, Fe,andZn,expressed on higher quality material than in ZKP96. As a consequence asǫ(X)=12+ log(X/H) whereX/His theabundanceofele- the microturbulence derived by RH99 (5.25 km s−1) is much mentX in numberrelativeto H. In thecase of RCrB andex- lower than the supersonic value found by ZKP99 (8 km s−1). tremeheliumstars,theabundancesarelistedasǫ(X)=12.15+ Since TPJ00 give a really unusual result we selected the data log(X/N),whereX/Nistheratioofnumberdensityofelement fromRH99. XtothetotalnumberdensityofnucleonsN.Notethatthesec- Object12:PG1704+222.TheanalysisofMH98is basedon onddefinitionismoregeneralthantheearlieroneandbothare better quality spectra than the preliminary results of CTM93, consistentiftheabundanceofheliumamountstoǫ(He)=11,a andtheirresultswereadopted. conditionwhichisfulfilledwithgoodaccuracyintheremain- Object14:IRAS18062+2410.Thereareemission linesvisi- ingstars(seeAsplundetal.2000andPandeyetal.2001).The bleinthespectrum.Thefourpaperscontainingadetermination 4 Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars Fig.1. a-(left)The ǫ(S) vs. ǫ(Zn) relation in our post-AGBsample. b-(right)The S/Zn vs. ǫ(Zn) relation. Objects from the R CrBclassandextremeheliumstarsarerepresentedbyblackcircles.Thenumberofobjectsinoursamplewithavailabledatais indicatedinthetoprightcornerofeachpanel. of atmosphericparametersare notindependent.The paper by and vWR00 are based on a higher quality spectrum than the AFGM01wassuggestedbytheworkofPGS00.Itisnotquite paperbyK95,yetthereisadifferenceintheabsolutedetermi- clear how the atmospheric parameters were determined. The nationofthecarbonabundance.Theremainingabundancesare analysis of RDM03 is based on non-LTE, fully blanketed at- similar.DatafromvWR00wereadopted. mosphericmodelsandthesamespectroscopicmaterialasused Object 37: IRAS06530−0213. The analysis of RvWG04 is in the LTE analysis by MRK02. Hence preferenceis givento based onhigherqualityspectroscopicmaterialthanHR03, so work by RDM03. Their values consistently point to a higher theirresultswereused. massofthestar. Object 50: IRAS12538−2611. The analysis of GAFP97 is Object 19: IRAS19475+3119. Since the paper of KPT02 is basedonhigherqualityspectraandmorelinesareusedforthe based on spectra with relatively low resolution (15000), the determinationofthechemicalabundances,sotheirvalueswere analysisbyAFGM01isprobablymorereliable. preferred. Object 20: IRAS17436+5003.Both analysis by KPT02 and Object52:BPSCS22877−0023.Bothanalysesleadtouncer- LBL90resultinamassivepost-AGBstar.KPT02spectrahave taindeterminationsofatmosphericparameters,butMH98used lowerresolutionthanthoseofLBL90,soweadoptedthedata alargernumberofmethods,soweadoptedtheirvalues. fromthelatter. Object54:LS3591.WeadoptedtheVSL98resultssinceonly Object 22: IRAS22223+4327. Both papers by vWR00 and theirworkgiveschemicalabundances. Object57:V453Ophandobject60:HD216457.Theabun- DvWW98arebasedonthesamespectroscopicalmaterialand danceanalysisofRDvW04 isbasedonhigherqualityspectra the same methods of analysis. We adopted the data from the andawiderspectralrangethanthatofGLG00andGLG98,so morerecentpapervWR00. weadoptedtheresultsfromRDvW04. Object 24: IRAS22272+5435. Both papers by ZKP95 and Object 98: IRAS15465+2818. The paper by AGL00 is the RLG02giveconsistentatmosphericparametersbutdifferinde- onlyonethatgiveschemicalabudances. rivedabundances,particularlyofFe.Otherdeterminationsare Object 113:V4732Sgr and object 117:FQ Aqr. The paper basedonrelativelyfewlines(C,N)oronjustoneline(Zn,O) of PRL01givesa consistentsetof abundancesforthe sample and maybe in error. Thepaper by RLG02 is based on higher ofextremeheliumstarsanalyzedhere,sotheirresultsarepre- qualitymaterialsoweusedtheirresults. ferred. Object 25: IRAS23304+6147The vWR00 analysis is based onmuchhigherresolutionspectra(60000)comparedtothatof KSP00(15000),soweusedtheresultsfromvWR00. 3. Empiricaldiagrams Object 29: IRAS04296+3429. The works by vWR00 and 3.1.Choiceofametallicityindicator DvWW98arebasedonthesamespectroscopicalmaterialand thesamemethodsofanalysis.DatafromvWR00wereadopted. Since it is expected that yields are strongly dependent on KSP99spectrahavelowerresolution. “metallicity” or, better said, on the initial chemical composi- Object35:IRAS07134+1005.AllanalysesofHD56126give tion ofthe star, we first haveto choosea reasonablemetallic- similarvaluesofatmosphericparameters.ThepapersbyHR03 ity indicator.As stressed, e.g.,by Mathis & Lamers(1992) or Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars 5 Lambert(2004),theusualmetallicityindicatorinstellaratmo- spheres,Fe,cannotbeusedforpost-AGBstarsbecauseofpos- siblestrongdepletionindustgrainsinaformerstageandsub- sequent ejection of the grains (dust-gas separation). Oxygen, themostabundant‘’metal”andthusthebesttheoreticalmetal- licity indicator,is possiblyaffected by nucleosynthesison the AGB(ONcycle,hotbottomburning).Fromthelistofelements forwhichwecompiledtheabundances,goodmetallicityindi- catorswouldbeSandZn.Thefirstoneisanα-element,likeO, anditsabundanceintheGalaxyisproportionaltothatofO.For the secondone,the nucleosynthesismechanismis unknowna priori, but Zn appears to roughly follow Fe (Mishenina et al. 2002).Figure1ashowsǫ(S)vs.ǫ(Zn)inourobjects.Thecor- relationisquitegood(thereisoneoutlier:VCrA, whichalso appearstobeasanextremeobjectinmanyoftheabundance- ratio diagrams of Asplund et al. 2000). The dispersion gives anideaofarealisticaverageabundanceuncertainty:about0.3 dex.Note that the data set spans a metallicity rangeof 2 dex. Figure1bshowsthatS/Znhasatendencytoincreaseasǫ(Zn) decreases (on average by 0.5dex per dex). This is the well- knownα-enhancementobservedinPopulationIIstars(Norris Fig.2. Theǫ(O)vs.ǫ(S)relation. etal.2001).Theeffectofα-enhancementonstellarparameters andstellarevolutioniscomplex(Kimetal.2002)andhasnot yetbeeninvestigatedinAGBstars.We choseSasourprinci- palmetallicityindicator,sincetherearemorestarswithdeter- minationsofSabundances(113)thanofZnabundances(76). We keepZnasasecondaryindicatorthatmighttesttheeffect ofα-enhancedmixturesinthestars.Wenote,however,thatthe Znabundancemeasurementsoftenrelyonlyononeline,which makestheevaluationofstatisticalerrorsdifficult. Figure 2 shows ǫ(O) vs. ǫ(S). The dispersion is such that obviouslyoxygencannotbechosenasametallicityindicatorin oursample.Thereasonforthisdispersionisnotclearapriori.It canbeduetonucleosynthesisandmixingaffectingtheoxygen abundance. But it could also be due to larger uncertainties in theoxygenabundancethanwasthought.NotethattheRCrB starsandextremeheliumstarsshowthelargestdispersionand thelargerproportionofobjectswithO/Ssmallerthansolar. 3.2.Nitrogenenhancement Figure3showslogN/Ovs.ǫ(O).TheobjectsfromtheRCrB Fig.3. ThelogN/Ovs.ǫ(O)relation. class and the extreme helium stars (which are represented by blackcircles)showdifferentbehaviourfromtherest:theydraw a clear anticorrelation between N/O and ǫ(O). This diagram He-burningproductshastobeinvokedtoreachthishighNen- hasalsobeenconstructedforplanetarynebulae(e.g.Henryet hancement. al.1989,Kingsburgh&Barlow1994,Leisy&Dennefeld1996) withdifferentresultsdependingontheauthorsandonthesam- 3.3.Indicationsofdredge-up ples.Someclaimnottoseeanyanticorrelation.Toourknowl- edge, never has the anticorrelation been seen so prominently Figure 4 shows ǫ(C+N+O) vs. ǫ(C). Only objects in which fora class ofPNe as forourR CrB and extremeheliumstars the abundancesof the three elements(C, N, andO) are avail- subsampleofpost-AGBstars.Oneinterpretationofsuchanan- able are represented here. This plot is very similar to the ticorrelation is the production of N at the expense of O (ON plot presented by Kingsburgh & Barlow(1994) and Leisy & cycle) brought to the stellar surface by the 2nd dredge-up. Dennefeld(1996)forplanetarynebulae,butwithalargernum- However, the N/O values reached by R CrB stars are much berofpoints.Theobjectswiththehighestcarbonabundances, higher than for the remaining post-AGB stars and for plane- which are mainly R CrB stars and extreme helium stars, are tarynebulae.Asplundetal.(2000)arguethatCNOcyclingon carbondominated.ThisagreeswithascenarioofCbeingpro- 6 Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars Fig.4. ǫ(C+N+O)vs.ǫ(C). Fig.6.[(C+N+O)/S](i.e.the logarithmof (C+N+O)/Sminus the logarithmof the solar value of this ratio) as a function of themassofthepost-AGBstar.Objectswithmasseslowerthan ducedbythetripleαreactionandbroughttothestarsurfaceby 0.55 M⊙ have been placed at 0.54 M⊙, objects with masses thethirddredge-up. largerthan0.94M⊙at0.95M⊙, 3.4.Dredge-upandstellarmass thanintheSun.Figure6plots[(C+N+O)/S](i.e.thelogarithm A more quantitative way to define whether dredge-up mech- of (C+N+O)/S minus the logarithm of the solar value of this anisms have occurred is to compare abundance ratios ob- ratio) as a function of the stellar mass. This diagram shows served in post-AGB stars or planetary nebulae with estimates that, according to our definition ([(C+N+O)/S] > 0), the vast of the initial abundance ratios. For example, Kingsburgh & majority of post-AGB stars in our sample (about 70%) have Barlow(1994) have proposed to call Type I PNe (understood experienced3rddredge-up.Italso suggeststhatthe massdis- as“objectsthathaveexperiencedenvelope-burningconversion tributionsofthetwosubclassesdonotdiffersignificantly. to nitrogen of dredged up primary carbon”) those objects in which the nitrogen abundance exceeds its progenitor’s C+N 3.5.Dredge-upandmetallicity abundance. As a proxy to the progenitor’s C+N abundance, theyusethevalueofC+NintheOrionnebula.Sincethepost- Theoreticalmodels(e.g.Marigo2001)predictthat3rddredge- AGB consideredhere are notnecessarily all close to the Sun, upismoreimportantatlowmetallicity.Figure7teststhispre- and because of abundance gradients in the Galaxy, perhaps a dictionbyplotting[(C+N+O)/S]asafunctionofǫ(S).Itshows safer way is to compare the N/S value in the post-AGB stars a net decrease in the efficiency of the 3rd dredge-up as the to the solar (C+N)/S value, instead of using abundanceswith metallicityincreases,inagreementwiththemodels.Thisisthe respectto hydrogen.Forthe solarabundances,we relyonthe first time that the observational evidence for this is so clear. compilation by Lodders(2003). Defining as Type I those ob- Note that, qualitatively, the same conclusion can be drawn, jects in which N/S is larger than the solar (C+N)/S ratio, we at least for bona-fide post-AGB stars, when using Zn instead show histograms of masses of Type I (left) and non-Type I of S as a metallicity indicator, as seen in Fig. 8, which plots (right)post-AGBstarsinFig.5.Itisseenthat,whileTypeIob- [(C+N+O)/S] as a function of ǫ(Zn). However, R CrB and jectsextendovertheentirerangeofmassesinoursample,more extreme helium stars tend to have higher [(C+N+O)/S] than thanhalfofthenon-TypeIobjectshavemassesbelow0.56M⊙. bona-fidepost-AGBstarsofsamemetallicity.Theyalsobehave The difference in stellar mass distributions is so tremendous differentlyinFigs.7and8,butmoreZnabundancedetermina- that it is highly significant, even taking into account the fact tionswouldbenecessarytomakethisclear. that error bars on stellar masses are large (as seen in Table 1 anddiscussedinSect.2.1).Theconclusionremainsthesame, 3.6.Carbon-richversusoxygen-richpost-AGBstars. althoughnotsostrong,ifweuseZninsteadofSasthemetal- licity indicator. However, when considering only R CrB and Figure9displaysthevalueofC/Oasafunctionofstellarmass. extremeheliumstars,nosuchdifferenceisseen. It clearly shows that oxygen-rich stars, at least in our sam- Similarly, one can identify objects that have experienced ple,tendtoaccumulateatthelowestmasses, whilethisisnot 3rddredge-upas those objectsin which (C+N+O)/Sis larger thecaseforcarbon-richstars.Notealsothattheproportionof Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars 7 Fig.5. Stellar mass distributionfor Type I post-AGBstars (left) and non-TypeI post-AGBstars (right).We call Type I those objectsinwhichN/Sishigherthanthesolar(C+N)/Svalue. Fig.7. [(C+N+O)/S] as a function of the metallicity as mea- Fig.8. [(C+N+O)/S] as a function of the metallicity as mea- suredbyǫ(S). suredbyǫ(Zn). 4. Summary,openquestions,andprospects carbon-richstars is 36%, while the proportionof stars having Themainaimofthispaperwastoshowtheutilityofpost-AGB experienced3rddredge-upisaslargeasabout70%.Ofcourse, starsto testtheoriesof AGBnucleosynthesis.So far,theonly these percentages should be taken with a grain of salt since testsofAGBnucleosynthesisbasedonlargesampleshavebeen the errors in determined abundances may place an object in madeusingplanetarynebulae.Post-AGBstarshaveseveralad- thewrongcategory.However,thedifferencebetweentheseper- vantagesoverplanetarynebulae:1) abundancesofa largeva- centagesandthelargespreadofC/O,aswellas[(C+N+O)/S] riety of elements can be derived, including of s-process ele- values,arguesinfavourofthenumberofcarbon-richstarsbe- ments;2)theabundanceofcarbon,anextremelyimportantel- ingsignificantlysmallerthanthenumberofstarshavingexpe- ement for the diagnostics, is known with the same accuracy rienced3rddredge-up.Qualitatively,thisisexpected,since3rd as the other elements, while in planetary nebulae, the carbon dredge-upisnotnecessarilysufficienttoproduceacarbonstar. abundance is significantly less reliable and more difficult to 8 Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars mer stages may have affected its present photospheric abun- dance).FollowingthedefinitionofTypeIplanetarynebulaeby Kingsburgh&Barlow (1994)butaccountingforthe metallic- ity,we definea classofTypeI post-AGBstars. We showthat non-typeIobjectsareinvastmajorityoflowmass(M⋆<0.56 M⊙).We alsoshowclearevidencethat3rddredge-upismore efficientatlowmetallicity. Wehavethusdemonstratedthepotentialofpost-AGBstars toconstrainthemodelsofAGBstarsandthepredictedyields. Thesampleofpost-AGBstarsislikelytogrowinthenearfu- ture, thanks to ASTRO-F, which is much more sensitive than IRASandshouldallowthediscoveryofmanyinfrared-excess stars amongwhich post-AGBstars are found.This will make the use of post-AGB stars even more attractive and powerful. In the present paper, we limited ourselves to only a few ele- mentsandtosimpleinterpretationswithoutdirectcomparison tomodels.Wedidnotaddressthequestionofobservationalbi- ases, which shouldbe investigatedby performingsimulations on models. Future studies will address other issues related to AGB nucleosynthesis,such as the productionof s-processel- Fig.9. Photospheric C/O versus the mass of the post-AGB ements,whicharemoreeasilydonewithpost-AGBstarsthan stars. withplanetarynebulae. Acknowledgements. This work was partly supported by grant obainthanthatofOandN;3)the determinationoftheatmo- 2.P03D.017.25ofthePolishStateCommitteeforScientificResearch spheric parameters T , and gravity g allows one to estimate and by the European Associated Laboratory ”Astronomy Poland- eff France”. G.S.isgrateful toBeatrizBarbuyandCorinneCharbonnel themassofthepost-AGBstarbycomparisonwiththeoretical forusefulcommentsonapreviousversionofthepaper.Commentsby stellarevolutionarytracks. thereferee,M.Reyniersaregratefullyacknowledged. Ofcourse,thestudyofpost-AGBstarshasitsowndifficul- ties.Inparticular,theabundanceanalysisisquitedifficultand manyeffectshavetobeconsideredindetail(seee.g.Asplund References et al. 2000). The lack of suitable lines for reliable analysis is Arellano Ferro, A.A., Giridhar, S., & Mathias, P. 2001, A&A, 368, oftenaproblem:i)forcoolerobjects,theoxygenabundanceis 250(AFGM01) hardtoderive,since usefullinesofoxygenonlystarttoshow Asplund, A., Gustafsson, B., Lambert, D.L., & Rao, N.K., 2000, up at temperaturesabove6000K;ii) nitrogen is often derived A&A,353,287(AGL00) from a few red lines, which are known to suffer non-LTE ef- Blo¨cker,T.,1995,A&A,299,755 Boothroyd,A.I.&Sackmann,I.-J.,1999,ApJ,510,232 fects;iii)starsthatarenotenrichedincarbon,sometimeshave Charbonnel,C.,2002,Ap&SS,281,161 onlyafewsuitablecarbonlines. Clayton,G.C.,1996,PASP,108,225 We considered all those objects from the present version Conlon, E.S., Dufton, P.L., & Keenan, F.P., 1994, A&A, 290, 897 of the catalogue of post-AGB objects (Szczerba et al.2001, (CDK94) Szczerba et al., in preparation)for which photosphericchem- Conlon,E.S.,Dufton,P.L.,Keenan,F.P.,&McCausland,R.J.H.,1991, ical abundances have been determined. We plotted diagrams MNRAS,248,820(CDK91) basedontheseabundances,similartotheonesbuiltforplane- Conlon, E.S., Theissen, A., Moehler, S., 1993, A&A, 269, L1 tarynebulaestudies.Thesametrendsasforplanetarynebulae (CTM93) werefound,butinaclearerfashion(e.g.N/Ovs.ǫ(O),reveal- Decin,L.,VanWinckel,H.,Waelkens,C.,&Bakker,E.J.,1998,A&A, ing the effect of the ON cycle, or (C+N+O)/H vs. ǫ(C) indi- 332,928(DvWW98) Drilling, J.S., Jeffery, C.S., & Heber, U., 1998, A&A, 329, 1019 catingthepresenceofobjectswithCproducedbythetripleα (DJH98) reaction).This is extremelyencouragingand shows the inter- ForestiniM.&Charbonnel,C.,1997,A&AS,123,241 estofusingpost-AGBstarstocomplementplanetarynebulae, Fujimoto,M.Y.,1977,PASJ,29,331 despitethedifficultiesinabundancedeterminations. Gauba,G.,&Parthasarathy,M.,2003,A&A,407,1007(GP03) Because thepost-AGBstars inoursampledonothaveall Gautschy,A.,Saio,H.,ARA&A1996,34,551 Giridhar, S., Arellano Ferro, A.A., & Parrao, L., 1997, PASP, 109, thesamemetallicities,wearguedthatabetterindicatorofthird 1077(GAFP97) dredge-upand/orhotbottomburningisobtainedbyconsider- Giridhar, S., Lambert, D.L., & Gonzalez, G., 1998, ApJ, 509, 366 ing the photosphericabundancesof C, N, O with respectto a (GLG98) metallicity indicator (and not with respect to H). It is there- Giridhar, S., Lambert, D.L., & Gonzalez, G., 2000, ApJ, 531, 521 fore these ratios that we compared with the solar ratios. We (GLG00) show that a convenient metallicity indicator is S (Fe cannot Giridhar, S., Rao, N.K., & Lambert, D.L., 1994, ApJ, 437, 476 be used for post-AGB objects because dust depletion in for- (GRL94) Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars 9 Gonzalez, G., Lambert, D.L., & Giridhar, S., 1997a, ApJ, 479, 427 Maas,T.,VanWinckel,H.,&LloydEvans,T.,2005,A&A,429,297 (GLG97a) (MvWLE05) Gonzalez, G., Lambert, D.L., & Giridhar, S., 1997b, ApJ, 481, 452 Maas, T., Van Winckel, H., & Waelkens, C., 2002, A&A, 386, 504 (GLG97b) (MvWW02) Gonzalez,G.,&Wallerstein,G.,1992,MNRAS,254,343(GW92) Marigo,P.,2001,A&A,370,194 Go´rny,S.K.Stasin´ska,G.,Tylenda,R.,1997,A&A,318,256 Marigo, P., Bernard-Salas, J., Pottasch, S.R., Tielens, A.G.G.M., Groenewegen,M.A.T.&deJong,T.1993,A&A,267,410 Wesselius,P.R.,2003,A&A,409,619 Hambly, N.C., Dufton, P.L., Keenan, F.P., & Lumsden, S.L., 1996, Marigo,P.,Bressan,A.&Chiosi,C.,1996,A&A,313,545 MNRAS,278,811(HDK96) Mathis,J.S.&Lamers,H.J.G.L.M.,1992,A&A,259,39 Harrison,P.M.,&Jeffery,C.S.,1997,A&A,323,177(HJ97) McCausland,R.J.H.,Conlon,E.S.,Dufton,P.L.,&Keenan,F.P.,1992, Heber,U.,1983,A&A,118,39(H83) ApJ,394,298(MCD92) Henry,R.B.C.,1990,ApJ,356,229 Mishenina,T.V.,Kovtyukh,V.V.,Soubiran,C.,Travaglio,C.,Busso, Henry,R.B.C.,Kwitter,K.B.&Bates,J.A.,2000,ApJ,531,928 M.,2002,A&A,396,189 Henry,R.B.C.,Liebert,J.,Boroson,T.A.,1989,ApJ,339,872 Moehler,S.,&Heber,U.,1998,A&A,335,985(MH98) Herwig, F., 2004, in Asymmetrical Planetary Nebulae III: Winds, Mooney,C.J.,Rolleston,W.R.J.,Keenan,F.P.,etal.,2002,MNRAS, StructureandtheThunderbird, ed.M.Meixner,J.H.Kastner,B. 337,851(MRK02) Balick&N.Soker,ASPConferenceProceedings,313,387 Munn,K.E.,Dufton,P.L.,Smartt,S.J.,&Hambly,N.C.,2004,A&A, Hrivnak,B.L.,&Reddy,B.,2003,ApJ,590,1049(HR03) 419,713(MDS04) IbenI.,Jr.&Truran,J.W.1978,ApJ,220,980 Napiwotzki, R., Heber, U., & Ko¨ppen, J., 1994, A&A, 292, 239 Izzard, R.G., Tout, C.A., Karakas, A.I., Pols, O.R., 2004, MNRAS, (NHK94) 350,407 Norris,J.E.,Ryan,S.G.,Beers,T.C.,2001,ApJ,561,1034 Jeffery,C.S.,1993,A&A,279,188(J93) Parthasarathy, M., Garc´ia-Lario, P., Sivarani, T., Manchado, A., & Jeffery,C.S.,&Heber,U.,1993,A&A,270,167(JH93) SanzFernana´ndezdeCo´rdoba,L.,2000,A&A,357,241(PGS00) Jeffery,C.S.,Hanill,P.J.,Harrison,P.M.,&Jeffer,S.V.,1998,A&A, Pereira,C.B.,Lorenz-Martins,S.,&Machado, M,2004,A&A,422, 340,476(JHHJ98) 637(PLM04) Jeffery,C.,&Heber,U.,1992,A&A,260,133(JH92) Pandey,G.,Lambert,L.D.,Rao,N.K.,&Jeffery,C.S.,2004,ApJ,602, Jeffery,C.S.,Hill,P.W.,&Heber,U.,1999,A&A,346,491(JHH99) L113(PLRJ04) Karakas,A.I.,Lattanzio,J.C.,Pols,O.R.,2002,PASA,19,515 Pandey,G.,Rao,N.K.,Lambert,L.D.,Jeffery,C.S.,&Asplund,M., Kendall, T.R., Dufton, P.L., Keenan, F.P., Beers, T.C., & Hambly, 2001,MNRAS,324,937(PRL01) N.C.,1997,A&A,317,82(KDK97) Pandey, G., Kameswara R.N., Lambert, D.L., Jeffery, C.S., & Kilkenny,D.,&Lydon,J.,1986,MNRAS,218,218(KL86) Asplund,M.2001,MNRAS,324,937 Kim,Y.-C.,Demarque,P.,Yi,S.K.,Alexander,D.R.,2002,ApJS,143, Peimbert,M.,1978,inPlanetaryNebulae,IAUSymposium78,p.217 499 Preston, G.W., Krzemi/’nski, W., Smak, J., & Williams, J.A., 1963, Kingsburgh,R.L.,Barlow,M.,1994,MNRAS,271,257 ApJ,137,401 Klochkova,V.G.,1995,MNRAS,272,710(K95) Rao, N.K., Giridhar, S., & Ashoka, B.N., 1990, MNRAS, 244, 29 Klochkova, V.G., & Panchuk, V.E., 1996, Bull. Spec. Astrophys. (RGA90) Observ.,41,5(KP96) Rao,N.K.,Goswami,A.,&Lambert,D.L.,2002,MNRAS,334,129 Klochkova, V.G., & Panchuk, V.E., & Tavolzhanskaya, N.S., 2002, (RGL02) Astron.Lett.,28,49(KPT02) Rao,N.K.,&Lambert,D.L.,2003,PASP,115,1304(RL03) Klochkova,V.G.,Szczerba,R.,&Panchuk,V.E.,2000,Astron.Lett., Reddy, B., Bakker, E.J., & Hrivnak, B.J., 1999, AJ, 117, 1834 26,88(KSP00a) (RBH99) Klochkova,V.G.,Szczerba,R.,&Panchuk,V.E.,2000,Astron.Lett., Reddy,B.,&Hrivnak,B.J.,1999,AJ,117,1834(RH99) 26,439(KSP00b) Reddy,B.,Lambert,D.L.,Gonzalez,G.,&Yong,D.,2002,ApJ,564, Klochkova, V.G., Szczerba, R., Panchuk, V.E., & Volk, K., 1999, 482(RLG02) A&A,345,905(KSP99) Klochkova, V.G., Yushkin, M.V., Miroshnichenko, A.S., Panchuk, Reddy, B., Parthasarathy, M., & Sivarani, T., 1996, A&A, 313, 191 V.E.&Bjorkman,K.S.,2002,A&A,392,143(KYM02) (RPS96) Kodaira,K.,1973,A&A,22,273 Renzini, A., 1979, in Stars and Star Systems, ed. B.E. Westerlund, Kwok, S., Volk, K., & Bidelman, W.P., 1997, ApJS, 112, 557 Reidel,Dordrecht,155 (KVB97) Renzini,A.&Voli,M.,1981,A&A,94,175 Iben,I.Jr,Tutukov,A.V.,&Yungelson,L.R.,1986,ApJ,456,750 Reyniers, M., Deroo, P., Van Winckel, H., Goriely, S., & Siess, L., Jura,M.,1986,ApJ,309,732 2004,Mem.S.A.It.,75,584(RDvW04) Lambert, D.L. in the XIII Canary Islands Winter School of Reyniers,M.,&VanWinckel,H.,2001,A&A,365,465(RvW01) Astrophysics, “Cosmochemistry. The melting pot of the ele- Reyniers, M., Van Winckel, H., Gallino, R., & Straniero, O., 2004, ments”, ed. C. Esteban, R. J. Garcia-Lo´pez, A. Herrero, F. A&A,417,269(RvWG04) Sa´nchez,CambridgeUniversityPress,p.81 Ryans,R.S.I.,Dufton,P.L.,Mooney,C.J.,etal.2003,A&A,401,1119 Lattanzio,J.,2002,NewAstronomyReviews,46,469 (RDM03) Leisy,P.&Dennefeld,M.,1996,A&AS,116,95 Scho¨nberner,D.,1983,ApJ,272,708 Lodders,K.,2003,ApJ,591,1220 Sivarani, T., Parthasarathy, M., Garc´ia-Lario, P., Manchado, A., & Luck,R.E.,&Bond,H.E.,1984,ApJ,279,729(LB84) Pottasch,S.R.,1999,A&AS,137,505(SPG99) Luck, R.E., Bond, H.E., & Lambert, D.L., 1990, ApJ, 357, 188 Stasin´ska, G., 2004, in the XIII Canary Islands Winter School of (LBL90) Astrophysics, “Cosmochemistry. The melting pot of the ele- Luck, R.E., Lambert, D.L., & Bond, H.E., 1983, PASP, 95, 413 ments”, ed. C. Esteban, R. J. Garcia-Lo´pez, A. Herrero, F. (LLB83) Sa´nchez,CambridgeUniversityPress,p.115 10 Stasin´skaetal.:Post-AGBstarsastestbedsofnucleosynthesisinABGstars Stasin´ska,G.,Go´rny,S.K.,Tylenda,R.,1997,A&A,327,736 Szczerba, R., Go´rny, S.K., Zalfresso-Jundziłło, 2001, in Post-AGB ObjectsasaPhaseofStellarEvolution,ed.R.SzczerbaandS.K. Go´rny,ASSLVol.265,p.13 Takeda,Y.,2003,A&A,402,343 The´venin,F.,Parthasarathy,M.,&Jasniewicz,G.,2000, A&A,359, 138(TPJ00) VanWinckel,H.,1995,Ph.D.Thesis,(vW95) VanWinckel,H.,1997,A&A,319,561(vW97) VanWinckel,H.,Oudmaijer,R.,D.,&Trams,N.,R.,1996,A&A,312, 553(vWOT96) VanWinckel,H.,&Reyniers,M.,2000,A&A,354,135(vWR00) VanWinckel,H.,Waelkens,C.,&Waters,L.B.F.M.,1996,A&A,306, L37(vWWW96) Vassiliadis,E.,Wood,P.R.,1994,ApJS,92,125 Venn, K.A.,Smartt,S.J.,Lennon, D.J.,& Dufton, P.L,1998, A&A, 334,987(VSL98) Ventura,P.,&D’Antona,F.,2005a,A&A431,279 Ventura,P.,&D’Antona,F.,2005b,astro-ph/0505221 Waelkens, C., Van Winckel, H., Bogaert, E., & Trams, N.R., 1991, A&A,251,495(WvWB91) Waelkens, C., Van Winckel, H., Trams, N.R., & Waters, L.B.F.M., 1992,A&A,256,L15(WvWTW92) WalkerH.J.,&Scho¨nbernerD.,1981,A&A,97,291(WS81) Zacˇs,L.,Klochkova,V.G.,&Panchuk,V.E.,1995,MNRAS,275,764 (ZKP95) Zacˇs, L., Klochkova, V.G., Panchuk, V.E., & Spe´lmanis, R., 1996, MNRAS,282,1171(ZKP96)